Accuracy evaluation of image registration and segmentation tools used in conformal treatment planning of prostate cancer

Impact of CT reconstruction algorithm on auto-segmentation performance

Model‐based iterative reconstruction (MBIR) reduces CT imaging dose while maintaining image quality. However, MBIR reduces noise while preserving edges which may impact intensity‐based tasks such as auto‐segmentation. This work evaluates the sensitivity of an auto‐contouring prostate atlas across multiple MBIR reconstruction protocols and benchmarks the results against filtered back projection (FBP). Images were created from raw projection data for 11 prostate cancer cases using FBP and nine different MBIR reconstructions (3 protocols/3 noise reduction levels) yielding 10 reconstructions/patient. Five bony structures, bladder, rectum, prostate, and seminal vesicles (SVs) were segmented using an auto‐segmentation pipeline that renders 3D binary masks for analysis. Performance was evaluated for volume percent difference (VPD) and Dice similarity coefficient (DSC), using FBP as the gold standard. Nonparametric Friedman tests plus post hoc all pairwise comparisons were employed to test for significant differences (P < 0.05) for soft tissue organs and protocol/level combinations. A physician performed qualitative grading of 396 MBIR contours across the prostate, bladder, SVs, and rectum in comparison to FBP using a six‐point scale. MBIR contours agreed with FBP for bony anatomy (DSC ≥ 0.98), bladder (DSC ≥ 0.94, VPD < 8.5%), and prostate (DSC = 0.94 ± 0.03, VPD = 4.50 ± 4.77% (range: 0.07–26.39%). Increased variability was observed for rectum (VPD = 7.50 ± 7.56% and DSC = 0.90 ± 0.08) and SVs (VPD and DSC of 8.23 ± 9.86% range (0.00–35.80%) and 0.87 ± 0.11, respectively). Over the all protocol/level comparisons, a significant difference was observed for the prostate VPD between BSPL1 and BSTL2 (adjusted P‐value = 0.039). Nevertheless, 300 of 396 (75.8%) of the four soft tissue structures using MBIR were graded as equivalent or better than FBP, suggesting that MBIR offered potential improvements in auto‐segmentation performance when compared to FBP. Future work may involve tuning organ‐specific MBIR parameters to further improve auto‐segmentation performance. Running title: Impact of CT Reconstruction Algorithm on Auto‐segmentation Performance.

Prostate contours delineation using interactive directional active contours model and parametric shape prior model

Prostate contours delineation on Magnetic Resonance (MR) images is a challenging and important task in medical imaging with applications of guiding biopsy, surgery and therapy. While a fully automated method is highly desired for this application, it can be a very difficult task due to the structure and surrounding tissues of the prostate gland. Traditional active contours-based delineation algorithms are typically quite successful for piecewise constant images. Nevertheless, when MR images have diffuse edges or multiple similar objects (e.g. bladder close to prostate) within close proximity, such approaches have proven to be unsuccessful. In order to mitigate these problems, we proposed a new framework for bi-stage contours delineation algorithm based on directional active contours (DAC) incorporating prior knowledge of the prostate shape. We first explicitly addressed the prostate contour delineation problem based on fast globally DAC that incorporates both statistical and parametric shape prior model. In doing so, we were able to exploit the global aspects of contour delineation problem by incorporating a user feedback in contours delineation process where it is shown that only a small amount of user input can sometimes resolve ambiguous scenarios raised by DAC. In addition, once the prostate contours have been delineated, a cost functional is designed to incorporate both user feedback interaction and the parametric shape prior model. Using data from publicly available prostate MR datasets, which includes several challenging clinical datasets, we highlighted the effectiveness and the capability of the proposed algorithm. Besides, the algorithm has been compared with several state-of-the-art methods.

Validation of the CT-MRI image registration with a dedicated phantom

PURPOSE

The present study was aimed at verifying the automatic registration of the Focal (Elekta) platform with a dedicated phantom.

MATERIALS AND METHODS

A phantom that simulates the pelvis region in a stylised way and finalised to the registration of computed tomography-magnetic resonance images was designed and realised. After acquiring the two sets of images, the registration was performed both in automatic and manual mode to verify whether they were comparable. To test the repeatability of the automatic registration, some known rigid transformations were imposed to the original images. If the registration method works correctly, parameters which bring the images into alignment must always be the same.

RESULTS

Automatic registration performed by the software did not prove satisfactory, whereas if a specific tool [volume of interest (VOI) tool] allowing the calculation to be limited to the landmark region was used, the registration parameters were comparable with those of the manual registration. Regarding the repeatability of the automatic registration, the software brought the images in the correct alignment performing translations and rotations along the longitudinal axis up to 40°, while it was not satisfactory for rotations along the transverse axes.

CONCLUSION

The experimental results showed that in clinical application automatic registration is reliable if the VOI tool that includes visible landmarks in both studies is used. However, because the algorithm did not prove sensitive to rotations along the transverse axes, the position of the patient during the examinations plays a crucial role.

Evaluation of the registration of temporal series of contrast-enhanced perfusion magnetic resonance 3D images of the liver

The registration of 2D and 3D images is one of the key tasks in medical image processing and analysis. Accurate registration is a crucial preprocessing step for many tasks; consequently, the evaluation of its accuracy becomes necessary. Unfortunately, this is a difficult task, especially when no golden pattern (true result) is available and when the signal values may have changed between successive images to be registered. This is the case this paper deals with: we have a series of 3D images, magnetic resonance images (MRI) of the liver and adjacent areas that have to be registered. They have been taken while a contrast is diffused through the liver tissue, so intensity of each observed point changes for two reasons: contrast diffusion/perfusion and deformation of the liver (due to body movement and breathing). In this paper, we introduce a new method to automatically compare two or more registration algorithms applied to the same case of a perfusion magnetic resonance dynamic image so that the best of them can be chosen when no ground truth is available. This is done by modeling the function that gives the intensity at a given point as a functional datum, and using statistical techniques to assess its change in comparison with other functions. An example of the application is shown by comparing two parametrizations of a B-spline based registration algorithm. The main result of the proposed method is a suggestive evidence to guide the physician in the process of selecting a registration algorithm, that recommends the algorithm of minimal complexity but still suitable for the case to be analyzed.

Assessment of accuracy and efficiency of atlas-based autosegmentation for prostate radiotherapy in a variety of clinical conditions

BACKGROUND AND PURPOSE

The goal of the current study was to evaluate the commercially available atlas-based autosegmentation software for clinical use in prostate radiotherapy. The accuracy was benchmarked against interobserver variability.

MATERIAL AND METHODS

A total of 20 planning computed tomographs (CTs) and 10 cone-beam CTs (CBCTs) were selected for prostate, rectum, and bladder delineation. The images varied regarding to individual (age, body mass index) and setup parameters (contrast agent, rectal balloon, implanted markers). Automatically created contours with ABAS(®) and iPlan(®) were compared to an expert's delineation by calculating the Dice similarity coefficient (DSC) and conformity index.

RESULTS

Demo-atlases of both systems showed different results for bladder (DSC(ABAS) 0.86 ± 0.17, DSC(iPlan) 0.51 ± 0.30) and prostate (DSC(ABAS) 0.71 ± 0.14, DSC(iPlan) 0.57 ± 0.19). Rectum delineation (DSC(ABAS) 0.78 ± 0.11, DSC(iPlan) 0.84 ± 0.08) demonstrated differences between the systems but better correlation of the automatically drawn volumes. ABAS(®) was closest to the interobserver benchmark. Autosegmentation with iPlan(®), ABAS(®) and manual segmentation took 0.5, 4 and 15-20 min, respectively. Automatic contouring on CBCT showed high dependence on image quality (DSC bladder 0.54, rectum 0.42, prostate 0.34).

CONCLUSION

For clinical routine, efforts are still necessary to either redesign algorithms implemented in autosegmentation or to optimize image quality for CBCT to guarantee required accuracy and time savings for adaptive radiotherapy.

[Automatic image segmentation for treatment planning in radiotherapy]

One drawback of the growth in conformal radiotherapy and image-guided radiotherapy is the increased time needed to define the volumes of interest. This also results in inter- and intra-observer variability. However, developments in computing and image processing have enabled these tasks to be partially or totally automated. This article will provide a detailed description of the main principles of image segmentation in radiotherapy, its applications and the most recent results in a clinical context.

Validation of mutual information-based registration of CT and bone SPECT images in dual-isotope studies

The registration of computed tomography (CT) and nuclear medicine (NM) images can substantially enhance patient diagnosis as it allows for the fusion of anatomical and functional information, as well as the attenuation correction of NM images. However, irrespective of the method used, registration accuracy depends heavily on the characteristics of the images that are registered and the degree of similarity between them. This poses a challenge for registering CT and NM images as they have very different characteristics and content. To address the particular problem of registering single photon emission computed tomography (SPECT) oncology studies with corresponding CT, we have proposed to perform a dual-isotope study with simultaneous injection of a tumor tracer and a bone imaging agent to obtain a tumor SPECT and a bone SPECT image that are inherently registered. As bone structures are generally visible in both CT and bone SPECT, performing registration of these images will be more easily attainable than registration of CT and tumor SPECT. By subsequently applying the spatial transformation determined from this registration to the tumor SPECT acquired from the same dual-isotope study, the optimal alignment between the CT and tumor SPECT images can be obtained. In this paper, we present the proof-of-concept of the proposed approach, the MI-based algorithm employed, and the techniques used to select the algorithm's parameters. Our objectives are to show the feasibility of CT and bone SPECT registration using this algorithm and to validate quantitatively the results generated using clinical data.

[Image registration for radiation therapy: Practical aspects and quality control]

The development of conformal radiotherapy techniques (CRT) and intensity modulated CRT requires an accurate delineation of target structures and organs at risk. Thus, additional information provided by anatomical and/or functional imaging modalities can be used for volume of interest determination combined with traditionally used Computed Tomography imaging (CT): for instance, functional or morphological Magnetic Resonance Imaging (f MRI or m MRI) or Positron Emission Tomography (PET). A prerequisite to the simultaneous use of this information is image registration. Due to the differences between the images and the information they provide, a quality control of image registration process for radiotherapy is mandatory. The purpose of this article is to present the difficulties in implementing such controls and to show the necessity for a clinical validation on patient's images. The last part of this work presents the possible interest in using f MRI to help radio-oncologists in the treatment planning for gliomas associated to image coregistration and quality control considerations.